Elon Musk’s Grok chatbot on Thursday started restricting the ability for non-paying users to create deepfake, sexualized images after global pushback from users and governments.
Late last month, some users began tagging Grok on Musk’s social media platform X with prompts such as “put her in a bikini” to generate nonconsensual images of real people on X.
The morphed images, including those of celebrities, politicians, and some minors, were created by Grok and posted publicly on X.
Before the current restrictions kicked in, Musk participated in the AI undressing trend by instructing Grok to create images of himself in a bikini.
Users and authorities complained and threatened to sue or punish X. Grok initially apologized for the inappropriate photos of what looked like children.
Now X has put Grok’s controversial morphing capabilities behind a paywall.
Governments, including the United Kingdom and the European Commission, said they aren’t satisfied with X’s safeguards as the ability to create and edit images is still available to paying users.
“While limiting AI image generation to paid users may marginally reduce volume and improve traceability, the abuse has not been stopped. It has simply been placed behind a paywall, allowing X to profit from harm,” said Emma Pickering, head of technology-facilitated abuse and economic empowerment at Refuge, in a statement.
As the public fury over what many have called AI-enabled harassment grows, regulators have taken notice, with some proposals to ban X in their countries.
Prime Minister of the U.K., Keir Starmer, said that X has “got to get a grip” of this “disgusting” trend. “We will take action on this because it’s simply not tolerable,” Starmer said.
In a veiled pushback, Musk appeared to criticize U.K. regulators outrage by reposting a claim that ChatGPT also permitted the creation of a bikini image of a non-real person through creative, non-explicit prompting.
After condemning Grok for producing sexualized images, the European Commission ordered X to retain all documents relating to Grok to ensure compliance with its rules.
“We need to be able to have access to them if we request it explicitly,” a spokesperson for The Commission said in a news conference.
France, India and Malaysia, and Brazil are also scrutinizing the platform.
While public figures were early targets for AI undressing, it soon tipped over to private individuals. On January 3, one user prompted Grok 50 times a day to generate nonconsensual, sexualized images of women in workplace settings, an analysis by AI-detection company Copyleaks.
Gender justice group Ultraviolet has mounted a campaign to pressure Apple and Google to remove X from their respective app stores on the grounds that X violates their child sexual abuse guidelines.
Musk has positioned his chatbot as a more edgy alternative compared to alternatives such as Gemini, ChatGPT and Claude, which have more restrictive guidelines around explicit content requests.
